The present invention relates generally to communication systems, both wired and wireless, employing a continuous phase modulation (“CPM”) waveform. One such CPM waveform is known as minimum shift keying (“MSK”) modulation. The present inventive system and method is applicable to all communication systems and radio frequency bands which utilize an MSK preamble, defined below, to determine baud rate, phase, frequency offset, and bit timing. More particularly, the inventive system and method is applicable to the military satellite communications UHF frequency band for deciding whether a signal of interest is present.
Many communication systems or networks, both wired (e.g., Ethernet) and wireless (e.g., HF, VHF, UHF radio), utilize a preamble to determine the modulation carrier frequency and phase. A MSK waveform with an alternating sequence, e.g., 1,1,0,0,1,1,0,0, . . . , has a characteristic frequency spectrum, sometimes referred to as the “MSK Tones” which also, in addition to carrier frequency and phase, provides modulation symbol rate and accurate baud timing of the MSK waveform. The preceding MSK alternating sequence may be written as [(1·2),(0·2)]m which may be generalized in the following form: [(1·n),(0·n)]m where the variable “n” may be referred to as the “symbol repetition factor” and the variable “m” may be referred to as the “symbol pair repetition factor”. Other MSK waveforms that fit this general pattern, e.g., 1,1,1,0,0,0,1,1,1,0,0,0, . . . , which can be written as [(1·3),(0·3)]m and 1,1,1,1,0,0,0,0,1,1,1,1,0,0,0,0, . . . , which can be written as [(1·40,(0·4)]m, are all contemplated by the present inventive system and method.
There are several problems with using an MSK waveform preamble for waveform detection and parameter estimation which must be overcome in order to allow for accurate measurement of symbol rate, frequency, phase, timing, and signal strength (measured, for example, as signal-to-noise ratio, carrier-to-interference ratio, etc.). In most communication systems it is highly desirable to limit the preamble time and begin to transmit information-carrying data as soon as possible since the preamble essentially displaces data-carrying capability. However, limiting the preamble time has the effect of limiting the signal energy of the preamble which increases the difficulty in finding the characteristic MSK tones in the frequency spectrum due to noise. The transition in energy at the start of the preamble exacerbates this problem because the unit step in energy at the start of the preamble induces tones in the frequency spectrum, making it harder to distinguish the MSK tones from the noise. This forces the acquisition algorithm to pay close attention to gain control because amplitude changes, which are usually implemented in gain “steps”, also add tones to the frequency spectrum. Additionally, tones are introduced to the frequency spectrum due to amplitude changes which are typically implemented in predetermined step changes by the gain control algorithm. Therefore, the gain control algorithm must attempt to minimize the number of step changes in order to minimize the introduction of tones into the frequency spectrum and thereby maximize the ability to detect the MSK tones.
The implementation of the Fourier Transform (“FT”) or the Laplace Transform (“LT”) are common methods to convert time domain data to frequency domain data for analysis. The FT has discrete bins which contain the energy as correlated with a frequency offset for each bin. The Fourier Transform may be implemented as a Discrete Fourier Transform (“DFT”) or as a Fast Fourier Transform (“FFT”) in those devices that are computationally-limited. It is to be understood that any method for converting time domain data to frequency domain data, such as Fourier Transforms, Laplace Transforms, discrete cosine, etc., are contemplated by the present invention and any method for converting time domain data to frequency domain data may be referred to herein, individually and/or collectively, as a Fourier Transform, or “FT” as would be consistent in the context as used.
The use of a FFT to find MSK tones at a specific spacing equal to one-quarter of the symbol rate of the received signal is described in the paper “An Innovative Synchronization Preamble for UHF MILSATCOM”, authored by Mark Miller, Mark A. Harris, and Donald R. Stephens (the “Miller Paper”), which is hereby incorporated herein by reference. This paper only describes the use of the FFT to find the MSK Tones and implies the use of a correlation function to determine the characteristic spacing of the MSK Tones. It describes the use of the FFT Bin number to find the Carrier Frequency and the Phase value of that center bin to find the Carrier Phase. It describes the use of the Phase difference between the carrier signal and the +/−1 MSK Tones to determine symbol phase (timing). However, use of the method disclosed in the Miller Paper results in a lot of false alarms, e.g., reporting a detection on noise, a foreign signal, or on an impulse signal. Part of the problem with the procedure used in the Miller paper is that the use of the center frequency of the bin with the largest amplitude is at best a rough estimate of the frequency of the carrier. The invention herein described includes the functionality of signal detection as well as a much more accurate method of signal parameter estimation. The present inventive system and method overcomes the problems of the prior art as detailed below.
Generally, there are two major issues with the FT approach to finding the characteristic MSK tones. The first problem deals with computational horsepower required to perform an FT. For example, a FT or DFT may be computationally burdensome, typically requiring N2 operations where N is the number of frequency bins. A FFT, by contrast, typically only requires N*log(N) operations and is therefore less computationally burdensome. However, with either an FT, DFT, or FFT, the fewer the number of bins the less frequency resolution is attainable. Therefore, an undesirable tradeoff is required between computational intensity and frequency resolution. Typical prior art solutions strike a compromise between computational intensity and frequency resolution by merely using the center frequency bin of the FT. The present inventive system and method overcomes the compromise problem by taking two contiguous FTs and, generally, determining the difference in phase for the bin of the center (carrier) frequency of the contiguous FTs to accurately determine the waveform frequency. The actual implementation used will be discussed in detail further below.
The second major issue with the FT implementation is that any frequency which is not an exact integer multiple of the data sample rate divided by the number of FT bins ends up with energy split between two adjacent bins of the FT. This effect tends to hide the characteristic MSK tones in the surrounding noise. The present inventive system and method solves this problem by multiplying the input data by a constant tone which may correspond to exactly ½ bin frequency, or some other fraction of bin frequency, and then performing another FT on the input data that has been multiplied by the constant tone and comparing the results with the results of the FT performed on the non-multiplied input data. The procedure creates two FT's on the same time-domain data. In one case, the carrier frequency will be located more closely to the center of an FT bin. In the other case, the carrier frequency will be located closer to the edge of the bin which may cause energy spillover into the next, adjacent FT bin. This energy spillover is undesired because it reduces the apparent signal strength in relation to the noise energy. The worst case situation occurs when tone appears exactly on a bin edge. In this case, the tone energy will be equally distributed between the two adjacent FT bins. For the single bin of interest (the transmitted carrier frequency), the apparent signal to noise ratio is 3 dB (a power ratio of 2) less than the signal to noise ratio which would be apparent if the FT was modified to locate the transmit carrier tone at the center of an FT bin. With two candidate FT's, choosing the transmit carrier frequency which has the highest magnitude will provide the best possible representation of the transmitted signal (in the frequency domain). This will improve the apparent signal to noise ratio which will improve the signal estimation performance in noisy environments.
The MSK preamble has an additional problem because a Fourier Transform of time domain data which is not an exact multiple of MSK symbols of the [(1·n),(0·n)]m pattern has an effect of varying the amplitude of the characteristic tones with respect to each other in a “walking” type pattern, i.e., the tone energy is constant but the tones repeated throughout the spectrum will not have a symmetric pattern in amplitude. This effect increases the difficulty in finding the characteristic (MSK) tones in the noise. To resolve the problem, the inventive system and method utilizes a polyphase resampler (interpolation and decimation method) to exactly place the sample rate of the time domain signal at an exact integer multiple of the symbol rate of the MSK waveform.
Thus there is a need for a system and method which can detect a continuous phase modulation waveform with a shortened MSK preamble and overcome the limitations of prior art systems/methods.
The present inventive system and method separates the detection and estimation functions. During detection, the baud rate is calculated as a first estimated parameter which is then utilized by the receiver for tailoring the signal sample rate and bandwidth to better match the incoming signal before calculating the transmitted carrier frequency, phase, and bit timing. The signal to noise ratio improvement which results from the tailoring process (of resampling and filtering) consequently improves the accuracy of the calculated values. The method described in the Miller paper does not separate the detection and estimation functions and therefore cannot accommodate the tailoring of the sample rate and bandwidth to match the incoming signal.
Additionally, the present inventive system and method performs a baud rate calculation by sorting the FT tones in order of amplitude and measuring the bin distance between the closest two tones. The expected value of the two closest tones is equal to the symbol rate divided by 2n where “n” is the number of bit repeats in the preamble from the [(1·n),(0·n)]m form of the MSK preamble. This baud rate calculation greatly improves the baud rate estimation algorithm as compared to the correlation described in the Miller paper because the inventive method takes advantage of the large signal to noise ratio apparent in a FT bin that contains a large signal level. The correlation method from the Miller paper integrates the noise in the entire range of frequencies which cover the transmit carrier frequency (e.g., +/−1500 Hz) and the MSK Tones (e.g., +/−28,000 Hz), which results in a sum of +/−29,500 Hz added into the correlation function. Additive White Gaussian Noise is, by definition, equally distributed in each frequency bin and the correlation process proposed in the Miller paper would integrate the noise in all of those bins. The sorting method used by the present invention excludes energy in those bins which do not contain enough signal energy to cause a signal detection. Therefore, the present invention excludes the noise from the baud rate estimation calculation as will become apparent in the detailed description below.
Furthermore, the detection calculation of the present invention determines the amplitude of largest tone (typically the carrier frequency) and adds the amplitudes of the +/−1 MSK Tones and the +/−2 MSK Tones in the signal detection calculation. The Miller paper does not address detection of the incoming signal at all.
The present invention also uses two adjacent FT windows to measure the phase difference of the carrier frequency between the two windows to thereby accurately determine the carrier frequency. Moreover, the present invention uses ½ tone spacing in a third FT window in order to place the amplitude of the carrier frequency in the most advantageous location so that carrier energy is not dispersed between two FT bins, as it would be if the carrier frequency were located at (or near) the edge of a bin. It is to be understood that the present invention contemplates offsetting the third FT by any amount, and is not limited in any way to just ½ tone spacing, so as to place the carrier frequency in the middle of the bin. The Miller paper is silent on the use of adjacent FT bins as well as on the use of a third FT spaced apart by a half tone.
One embodiment of the present invention avoids the problems of the prior art by using two or more contiguous Fourier Transforms for detecting a continuous phase modulation waveform and determine the characteristics of the waveform such as frequency, phase, timing, and signal strength.
Accordingly, it is an object of the present invention to obviate many of the above problems in the prior art and to provide a novel system and method for detecting a continuous phase modulation waveform.
It is another object of the present invention to provide a novel system and method for determining the waveform characteristics of a continuous phase modulation waveform.
It is yet another object of the present invention to provide a novel system and method for using two or more Fourier Transforms in the detection of a continuous phase modulation waveform with an MSK preamble.
It is still another object of the present invention to provide a novel system and method for using two or more Fourier Transforms to determine the characteristics of a received continuous phase modulation waveform with an MSK preamble.
It is a further object of the present invention to provide a novel system and method for using two or more Fourier Transforms for acquiring a continuous phase modulation waveform with an MSK preamble.
It is yet a further object of the present invention to provide a novel system and method of determining the frequency of a received and detected continuous wave waveform.
These and many other objects and advantages of the present invention will be readily apparent to one skilled in the art to which the invention pertains from a perusal of the claims, the appended drawings, and the following detailed description of the preferred embodiments.